IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 54, NO. 9, SEPTEMBER 2007
825
Fast-Transient PCCM Switching Converter With
Freewheel Switching Control
Dongsheng Ma, Senior Member, IEEE, and Wing-Hung Ki, Member, IEEE
Abstract—This brief presents a new switching converter operating in pseudo-continuous-conduction mode (PCCM) with freewheel switching control. Compared with conventional discontinuous-conduction mode (DCM) converters, this converter demonstrates much improved current handling capability with reduced
current and voltage ripples. The control-to-output transfer function exhibits a single-pole behavior, making the load transient response much faster than its CCM counterparts. Simulation and
experimental results show that, with a 6-V, 6-W load and a 10-V
unregulated supply, the PCCM converter has a current ripple of
only 1.1 A and a ripple voltage of only 58 mV, while a DCM converter has a current ripple of 2.2 A and a ripple voltage of 220 mV.
In addition, the PCCM converter takes only 25 s to respond to a
500-mA load current change while a CCM one requires 1.4 ms.
Index Terms—Current ripple, freewheel switching, pseudo-continuous-conduction mode (PCCM), switching converter, transient
response, voltage ripple.
I. INTRODUCTION
D
EVELOPMENT of high-performance electronic devices
imposes new challenges on power supplies. With the
supply voltage reduceing from 5 to 3.3, 2.5, and even 0.8 V in
the future, the current level in state-of-the-art computer CPUs
increases from a few amperes to 30–50 A for faster computation and more powerful functions [1]. These large numbers
of computing activities lead to drastic current changes in a
short time in the power supplies. To maintain high regulation
performance and to ensure system stability, the power supplies
must have the capability of handling large output current with
fast transient response [2].
Switch mode power converters, or switching converters in
short, have been widely. It is an essential component to provide clean and quality power for low-voltage electronic devices.
Fig. 1 shows the circuit block diagram of a conventional buck
switching converter. It provides electronic devices with regulated supply voltage, high efficiency and flexible voltage conversion. Compared to boost and flyback topologies, the buck
converter enjoys a continuous output current when operating in
continuous-conduction mode (CCM), leading to smaller current
ripple and thus lower switching noise. In addition, CCM operation is much more favorable for large current applications, because it can deliver more current than the converter operating in
discontinuous-conduction mode (DCM) [3]. However, a DCM
converter usually has a much faster transient response and a loop
gain that is easier to compensate than a CCM converter. With the
Manuscript received November 29, 2006; revised March 9, 2007. This paper
was recommended by Associate Editor E. Alarcon.
D. Ma is with the Department of Electrical and Computer Engineering, the
University of Arizona, Tucson, AZ 85721 USA (e-mail: ma@ece.arizona.edu).
W.-H. Ki is with the Department of Electronic and Computer Engineering,
the Hong Kong University of Science and Technology, Hong Kong.
Digital Object Identifier 10.1109/TCSII.2007.900903
Fig. 1. Conventional buck switching converter.
great demands on both fast transient response and large current
capability, it would be very desirable to have a new converter
incorporating advantages of both CCM and DCM converters.
To solve the mentioned issues, we present in this brief a converter that operates in a new operation mode—the pseudo CCM
(PCCM), made possible by freewheel switching control. The
rest of the brief is organized as follows. In Section II, we review
state-of-the-art switching converter designs, and then discuss
their drawbacks for high-current fast-transient applications. In
Section III, we introduce our design with detailed circuit structure, operation scheme and topology extensions. Simulation and
experimental results are provided in Section IV. Finally, we conclude our research efforts in Section V.
II. REVIEW OF PRIOR ARTS & DESIGN MOTIVATION
A. Current Ripple and Switching Noise at Heavy Load
The operation of a switching converter can be classified as
either CCM or DCM, according to its inductor current, as shown
in Fig. 2. When the load current is heavy ( in Fig. 1 is small),
the inductor current stays above zero to supply more current to
the output. The converter thus operates in CCM [Fig. 2(a)], and
each switching period
is divided into two parts:
and
. It ramps down during
when
is
off and is on. When the load is light ( in Fig. 1 is large), the
inductor current ramps up, then down to and stays at zero every
cycle. Each switching cycle is divided into three parts:
,
and
. The inductor current
waveform in
and
is similar to that in CCM, but
stays at zero during
when and are off. The converter
then operates in DCM [Fig. 2(b)]. For a CCM converter, the dc
level of the inductor current
is increased to supply a larger
load current, while the current ripple
is kept constant and
small. Hence, the output ripple voltage and switching noise are
relatively small. However, for a DCM converter, to supply the
same load current, a smaller value of inductor has to be used,
and the inductor current increases more sharply from zero to a
1549-7747/$25.00 © 2007 IEEE
826
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 54, NO. 9, SEPTEMBER 2007
Fig. 2. Inductor current of a switching converter in (a) CCM and (b) DCM.
much higher peak current
[Fig. 2(b)]. As a result, the current
is very large, generating large switching noise and
ripple
ripple voltages at the output.
B. Transient Response and Loop Gain Bandwidth
A switching converter may operate with either voltage- or
current-programming control. With a voltage-programming
control, the loop gain transfer function can be shown to be [4]
(1)
where
is the gain of the error amplifier that consists of an
op-amp and a compensation network,
is the control-tooutput transfer function, is the scaling factor,
is the duty
ratio of the converter,
is the peak-to-peak voltage of the oscillator ramp,
is the output voltage,
is the
frequency of the complex poles, and
is the quality factor.
As shown in Fig. 3(a), the existence of the complex poles in
the control-to-output transfer function makes it very difficult to
design a compensator to achieve a loop gain with large bandwidth [4]. When the converter operates in current-programming
control, the control-to-output transfer function consists of two
separated real poles [4]. Pole–zero compensation methods can
then be used to achieve a larger loop-gain bandwidth [5]. However, a current sensing circuit is needed for loop control and special ramp compensation technique is needed to avoid sub-harmonic oscillation [6]. As a result, the operation of the converter
is very sensitive to the current sensing signal that is seriously
corrupted by switching noise. The corresponding circuit design
complexity would be high as well. Hence, our research efforts
are put on voltage-programming converters in this brief.
As discussed in Section II-A, although DCM operation is
not preferred for heavy load current applications, it exhibits a
very attractive feature on the transient response. This can be explained by its loop gain transfer function [7]
(2)
where
is the voltage conversion ratio, R is
the equivalent load resistance,
and
is the
Fig. 3. Bode plots of the control-to-output transfer function, when a buck converter operates in (a) CCM, and (b) DCM.
switching period. For the control-to-output transfer function
of a DCM buck converter, there is only one pole in the low
frequency range as shown in Fig. 3(b) [8], and the converter
can be stabilized by pole–zero compensation, giving a much
broader bandwidth than the CCM converter [9].
III. PROPOSED PCCM CONVERTER
A. Proposed Converter With Freewheel Switching Control
To effectively reduce the peak inductor current in single-inductor multiple-output (SIMO) switching converters, we proposed a freewheel switching control scheme [10]. Due to the
nature of time-multiplexing control between the sub-converters,
the transient performance was not much improved. However, if
the technique is applied to traditional single-output converters,
both current handling capability and transient response can be
significantly enhanced.
A new switching converter is thus proposed in Fig. 4(a). The
operation can be explained with reference to Fig. 4(b). Different
from conventional buck converters, an additional pMOS power
(named a freewheel switch) is added across the inswitch
, the pMOS power switch
is turned on,
ductor. During
and
are both off. The inductor curwhile the switches
. In
,
and
rent increases with a slope of
are turned off. The nMOS power switches
is turned on.
until it
The inductor current decreases with a slope of
hits a current level of . Then the converter enters the period of
MA AND KI: FAST-TRANSIENT PCCM SWITCHING CONVERTER WITH FREEWHEEL SWITCHING CONTROL
827
Fig. 4. (a) Block diagram of the proposed converter. (b) Timing diagram of the
proposed converter.
Fig. 6. (a) Pole–zero compensation for PCCM buck converter. (b) Dominantpole compensation for CCM buck converter.
Fig. 5. Inductor current waveforms in three different operation modes.
where is turned on while both and are turned off.
because the switch
The inductor current stays constant at
shorts the inductor L and the voltage across the inductor is thus
equal to zero. The operation of freewheel switching buck converter is similar to a traditional DCM one, but the inductor curinstead of zero during
.
rent stays at a constant value of
This allows the converter to deliver larger current by simply
in Fig. 5.
boosting the current level of
is adaptively determined by the real-time load
The level of
is fixed, the load power is linpower of the converter. Since
early proportional to the load current and the inductor current. A
compact current sensing circuit is thus implemented for online
current measurement [11]. Based on the load current, one of the
will be used. If the current sensing
eight predefined levels of
is chosen.
circuit senses a higher current, a higher level of
Otherwise, a lower one would be used for a low power load. The
is always slightly lower than
control circuit makes sure that
(Fig. 5) to ensure PCCM operation,
the average load current
accordingly to limit the turn-on time of
to be
and adjusts
less than 5% of each switching period. Hence, the average curis much less than that of
and , and the power
rent of
loss due to
can be very low, allowing
to be implemented
by a much smaller power transistor. This operation mode was
named the PCCM in [10], because it achieves a similar current
handling ability as a CCM converter, but the inductor waveform
resembles a DCM converter. The advantages of the converter
are reduced current and voltage ripples.
B. Bandwidth Enhancement on the Proposed Design
Another feature of PCCM operation is a better transient performance. Similar to the DCM converter, the PCCM converter
exhibits a single-pole system behavior. This can be explained
with reference to Fig. 2(b). For a CCM converter, the power
stage is a second-order system due to the two reactive components and with corresponding state variables the capacitor
and the inductor current . However, for a DCM
voltage
is reset to zero every switching cycle, eliminating
converter,
as a state variable and the order of the system is reduced
is reset to a constant
every
to 1. For the PCCM case,
is equivaswitching cycle. In terms of small-signal analysis,
lent to reset to zero, and the transfer function of the power stage
is then first order. Therefore, pole–zero compensation [9] can be
use to broaden the loop-gain bandwidth, as shown in Fig. 6(a).
For comparison, the dominant pole compensation method for a
CCM converter is demonstrated in Fig. 6(b).
828
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 54, NO. 9, SEPTEMBER 2007
Fig. 7. PCCM converters with freewheel switch in (a) buck, (b) boost, (c) flyback, and (d) non-inverting flyback topologies, respectively.
TABLE I
DESIGN SPECIFICATIONS OF THE THREE CONVERTERS
Fig. 9. Output voltage and inductor current of CCM converter.
Fig. 10. Output voltage and inductor current of PCCM converter.
Fig. 8. Output voltages and inductor currents of the three converters.
C. Topology Extensions
The addition of the freewheel switch
across the inductor
for PCCM operation could be extended to other converter
topologies, as shown in Fig. 7.
IV. SIMULATION AND EXPERIMENTAL RESULTS
To verify the design ideas, three buck converters, operating
in DCM, CCM, and PCCM, respectively, are designed. The key
design specifications are listed in Table I. For fair comparison,
the converters operate with the same power load, switching frequency and input and output voltages. However, to deliver the
same power, the change in the inductor current of the DCM converter has to be much larger than the others, and a smaller inductor (10 H) has to be used. Consequently, to keep a similar
ripple voltage, the DCM converter needs a larger filtering capacitor (50 F). Despise a larger capacitor, the DCM converter still
resulted in the largest current and voltage ripples (Fig. 8) and it
also causes a sluggish transient response. Hence, the DCM converter gives the worst transient response and current and voltage
ripples. The rest of the comparisons are thus focused on the
CCM and PCCM converters.
Figs. 9 and 10 show the measured output voltage and inductor
current of the CCM and PCCM converters, respectively. For the
CCM converter, the peak-to-peak ripple voltage is 48 mV with
52-mV glitches, and the measured current ripple is 1.2 A. For
the PCCM converter, the peak-to-peak ripple voltage is 58 mV,
with 41-mV glitches, and the measured current ripple is 1.1 A.
Hence, with the same load power of 6 W, the PCCM converter
has similar current handling capability as the CCM converter
without compromising voltage and current ripples. To evaluate
the transient performance, we first examine the compensation
methods used by the converters. As discussed in Section II-B,
the CCM converter has a pair of complex poles. To guarantee a
stable operation, dominant pole compensation is used to reduce
the gain below 0 dB before reaching the resonance frequency
of the complex poles. With the parameters in Table I, the loop
gain unity gain bandwidth is designed to be 700 Hz, as shown in
Fig. 11. Even with a low bandwidth, the gain margin is only 9 dB
at the complex-pole frequency of 8 kHz. Dominant compensation can also be used for the PCCM converter. However, since
there is only one low-frequency pole in the control-to-output
transfer function, pole–zero compensation could be used to extend the bandwidth [8]. The introduced zero is located right at
the low frequency pole in the control-to-output transfer function. As shown in Fig. 12, the bandwidth of the loop gain for the
proposed PCCM converter is 17 kHz, which is 24 times larger
than that of the CCM converter.
MA AND KI: FAST-TRANSIENT PCCM SWITCHING CONVERTER WITH FREEWHEEL SWITCHING CONTROL
829
Fig. 11. Loop gain of the CCM converter.
Fig. 14. Load transient performance of the PCCM converter.
mV. In Fig. 14, for the PCCM converter, the measured transient response to a much faster load change is only 25 s, with
a peak-to-peak voltage variation of 185 mV.
V. CONCLUSION
Fig. 12. Loop gain of the PCCM converter.
A pseudo CCM buck converter with freewheel switching control is proposed in this brief. The new converter inherits the
advantages of both CCM and DCM converters: it can handle
large current stress while maintaining low current ripple and
switching noise. The control-to-output part of the converter only
introduces one single low-frequency pole. Pole–zero compensation can thus be used to achieve a faster transient performance.
Simulation and experimental results successfully verified the
merits of the proposed design.
REFERENCES
Fig. 13. Load transient performance of the CCM converter.
Load transient measurements were then performed. For each
of the converters, we periodically switched the load between 3
and 6 W. Due to different response times, the load of the PCCM
converter was switched much quicker than that of the CCM converter. Figs. 13 and 14 show the measured results of the CCM
and the PCCM converter, respectively. In each figure, the upper
waveform shows the output voltage of the converter, while the
lower one shows the load current. In Fig. 13, for the CCM converter, the measured transient response to the described load
change is 1.4 ms, with a peak-to-peak voltage variation of 310
[1] International Technology Roadmap for Semiconductors, 2003 Edition,
2003 [Online]. Available: http://www.public.itrs.net/
[2] M. T. Zhang et al., “Design consign considerations for low-voltage
on-board DC/DC modules for next generations of data processing circuits,” IEEE Trans. Power Electron., vol. 11, no. 3, pp. 328–337, Mar.
1996.
[3] R. Erickson and D. Maksimovic, Fundamentals of Power Electronics,
2nd ed. Norwell, MA: Kluwer, 2000.
[4] W.-H. Ki, “Signal flow graph in loop gain analysis of dc–dc PWM
CCM siwtching converters,” IEEE Trans. Circuits Syst. I, Fundam.
Theory Appl., no. 6, pp. 644–654, Jun. 1998.
[5] C. Lee and P. Mok, “A monolithic current-mode CMOS dc–dc converter with on-chip current sensing technique,” IEEE J.Solid-State Circuits, vol. 39, no. 1, pp. 3–14, Jan. 2004.
[6] W.-H. Ki, “Analysis of subharmonic oscillation of fixed-frequency current-programming switch mode power converters,” IEEE Trans. Circuits Syst. I, Fundam. Theory Appl., vol. 45, no. 1, pp. 104–108, Jan.
1998.
[7] V. Tam, “Loop gain simulation and measurement of PWM switching
Converters,” M.Phil thesis, HKUST, Hong Kong, 1999.
[8] J. Sun et al., “Modeling of PWM converters in discontinous-conduction
mode—A reexamination,” in Proc. IEEE Power Electron. Specialists
Conf., 1998, vol. 1, pp. 615–622.
[9] D. Ma et al., “Single-inductor dual-output CMOS switching converters
in discontinuous-conduction mode with time-multiplexing control,”
IEEE J.Solid-State Circuits, vol. 38, no. 1, pp. 89–100, Jan. 2003.
[10] D. Ma, W.-H. Ki, and C. Y. Tsui, “A pseudo-CCM/DCM SIMO
switching converter with freewheel switching,” IEEE J. Solid-State
Circuits, vol. 38, no. 6, pp. 1007–1014, Jun. 2003.
[11] D. Ma and C. Zhang, “Thermal compensation method for CMOS
digital integrated circuits using temperature-adaptive dc–dc converter,” IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 53, no. 11, pp.
1284–1288, Nov. 2006.